The hair cell analysis toolbox is a precise and fully automated pipeline for whole cochlea hair cell quantification
Fig 5
HCAT detection performance on published images of cochlear hair cells.
HCAT detection performance was assessed by running a cell detection analysis in the GUI on 88 confocal images of cochlear hair cells sampled from published figures across 15 different original studies [16–30]. None of the images from this analysis were used to train the model. Each image was adjusted within the GUI for optimal detection performance. Cells in each image were also manually counted (presented as ground truth values) and results compared to HCAT’s automated detection. The resulting population distributions of hair cells are compared for OHCs (A), and IHCs (B). The mean difference in predicted number of IHCs (open circles) and OHCs (filled circles) in each publication is summarized for each cell type: zero indicates an accurate detection, negative values indicate false negative detections, while positive values indicate false positive detections (C). GUI, graphical user interface; HCAT, hair cell analysis toolbox; IHC, inner hair cell; OHC, outer hair cell.